Use frequent terms and reading statistics as a document overview
Compare draft vocabulary, estimate reading effort or choose documents to inspect first. This tool creates a Markdown report of word counts, sentence segments, characters, estimated reading time and recurring words or word pairs.
Frequency is not the same as importance. A repeated company name can outrank a one-sentence conclusion, and a frequently repeated warning does not mean the warning applies everywhere. The tool uses local text rules rather than a language model: it does not interpret meaning, classify sentiment or measure search demand for the reported terms.
Interpret each number according to how it is produced
| Your situation | Recommended approach |
|---|---|
| Estimate time for a first reading | Use reading time as a rough baseline of 200 words per minute, rounded up with a minimum of one minute. Add time for equations, tables, unfamiliar vocabulary and note taking. |
| Identify recurring vocabulary | Review the ranked keywords against headings and examples. Common English stop words, short tokens and number-only tokens are excluded from keyword ranking. |
| Compare two versions of a document | Use the same extraction and OCR settings for both. Differences can reflect changed headers, text layers or scan recognition as well as actual edits. |
| Research SEO keywords | Use this only to inspect a document's wording. The report contains no search volume, competition, trend or ranking data. |
Check vocabulary and the reading-time calculation
The research-notes.pdf sample has two digital sections that discuss browser privacy and document checks. Its repeated vocabulary gives the frequency analysis something concrete to report.
- Choose research-notes.pdf and uncheck Use local English OCR when a scanned page has no readable text.
- Select Find keywords and reading time. Review the Keywords and reading statistics text, including the Words and Estimated reading time entries.
- Divide the reported word count by 200, round upward to a whole minute, and use one minute if the result would be smaller. Compare your calculation with the report.
- Download insights and compare several suggested terms with their source sentences. Record which terms describe the topic and which are merely repeated wording.
Expected result. The Markdown report contains positive word and character counts, a bounded keyword list and the stated reading-time estimate. Suggested phrases reflect recurring filtered tokens; they are not generated summaries or measured search queries.
Check the downloaded result
- Read any notice that pages were not OCR-processed. An analysis of the available text can succeed even when a scanned portion contributes no useful text.
- Search the original document for a reported phrase before quoting it. Pairs are formed after stop-word filtering, so their words need not have appeared directly next to one another in the source.
- Compare the report with Extract PDF text if the word count surprises you. Running headers and footers can be suppressed, while reading order and OCR substitutions can change the extracted text.
- Treat the Sentences value as a filtered analysis count. Very short and very long segments are excluded; it is not suitable as an exact editorial sentence total.
Troubleshooting
- The keywords omit a short but important term
Keyword candidates shorter than three characters, number-only tokens and common English stop words are removed. Ranking also favors recurring two-word combinations.
Try this: Add essential acronyms or identifiers to your own notes after checking the source. Do not infer that an omitted term is absent or unimportant.
- A keyword phrase does not appear as an exact quotation
Word pairs are formed from the filtered token sequence, which can remove connecting words or cross original line boundaries.
Try this: Search the individual words and read the surrounding sentence. Describe the result as a vocabulary clue, not a verbatim phrase from the author.
- Reading time seems too short for a technical document
The formula considers word count only. It does not add time for charts, mathematical derivations, comprehension or the reader's language proficiency.
Try this: Use your own reading pace and add time for non-prose content. The displayed estimate is a consistent baseline, not a timed assessment.
- Words or counts differ from another editor
Tools use different token boundaries, header removal and sentence rules. OCR and broken PDF text encodings introduce further differences.
Try this: Inspect extracted text and compare the same portion of the source in both tools. Use an agreed counting method when a submission limit depends on the result.
Limits and compatibility
- Counting rules
- Reading time includes all recognized word tokens. Keyword ranking is English-focused and returns at most 18 entries. Sentence analysis retains segments of 20–800 characters and at most 4,000 segments.
- Language and local analysis
- The browser processes available text locally. English stop-word rules are not a multilingual language analysis, and optional English OCR cannot guarantee transcription accuracy.
Your files and form values are processed on this device. Downloaded copies remain wherever you save them. See how to check local processing and our privacy policy for site data and optional services.
Common questions
- Does a higher-ranked word mean the author agrees with it?
- No. The analysis does not distinguish endorsement, criticism, quotation or negation. Read the surrounding passage to understand how the term is being used.
- Why does a very short PDF still show one minute?
- The estimate has a minimum of one minute and rounds upward. It is a coarse planning figure rather than a stopwatch measurement.
Continue your document workflow
Maintained by Shakti Prasad Ram. Found a mismatch? Report the tool, browser and steps without sending a sensitive document. Back to the tool